CN109887234B

CN109887234B - Method and device for preventing children from getting lost, electronic equipment and storage medium

Info

Publication number: CN109887234B
Application number: CN201910172200.8A
Authority: CN
Inventors: 杨尊程
Original assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Current assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Priority date: 2019-03-07
Filing date: 2019-03-07
Publication date: 2023-04-18
Anticipated expiration: 2039-03-07
Also published as: CN109887234A

Abstract

The invention discloses a method and a device for preventing children from getting lost, electronic equipment and a storage medium, wherein the method comprises the following steps: acquiring a monitoring video stream of a target scene; determining whether a child is present in the target scene based on the surveillance video stream; if the children exist in the target scene, performing depth analysis on the monitoring video stream to obtain an analysis result; determining whether the child is at risk of getting lost based on the analysis result; and if the child has the risk of losing, giving an alarm. By adopting the technical scheme, the purpose of monitoring all children in a large-area range in a loss prevention manner is achieved.

Description

Method and device for preventing children from getting lost, electronic equipment and storage medium

Technical Field

The embodiment of the invention relates to a child protection technology, in particular to a method and a device for preventing children from getting lost, electronic equipment and a storage medium.

Background

In places with dense crowds such as large markets and downtown areas, children are easy to lose due to the dense crowds, crowds and the like.

At present, the commonly adopted scheme for preventing children from getting lost is as follows: through wearing the positioning device for children on one's body, this positioning device is used for fixing a position children to carry out information interaction with children guardians's intelligent terminal in real time, in case detect children and its guardians between the distance when far away, then report to the police.

In the above-mentioned scheme of preventing children from wandering away, need purchase in advance and wear for every children positioning device, the cost is higher, and when discovering children and walk away, can't in time keep children to wander away the information on scene, in case children fall into criminal's hand, children wear positioning device can be abandoned and destroyed. Moreover, for scenes with dense crowds such as superstores, downtown areas and the like, it cannot be guaranteed that each child wears the positioning device, and the purpose of preventing the child from being lost cannot be achieved for the children who do not wear the positioning device.

Disclosure of Invention

The invention provides a method and a device for preventing children from getting lost, electronic equipment and a storage medium, and achieves the purpose of preventing all children in a large-area range from getting lost and monitoring.

In a first aspect, an embodiment of the present invention provides a method for preventing a child from getting lost, including:

acquiring a monitoring video stream of a target scene;

determining whether a child is present in the target scene based on the surveillance video stream;

if the children exist in the target scene, performing depth analysis on the monitoring video stream to obtain an analysis result;

determining whether the child is at risk of getting lost based on the analysis results;

and if the child has the risk of losing, giving an alarm.

In a second aspect, embodiments of the present invention further provide a device for preventing a child from getting lost, the device including:

the acquisition module is used for acquiring a monitoring video stream of a target scene;

a child determination module, configured to determine whether a child exists in the target scene based on the surveillance video stream;

the analysis module is used for carrying out depth analysis on the monitoring video stream to obtain an analysis result if the children exist in the target scene;

a risk determination module for determining whether the child is at risk of getting lost based on the analysis result;

and the alarm module is used for giving an alarm if the child has the risk of losing.

In a third aspect, an embodiment of the present invention further provides an electronic device, where the electronic device includes:

one or more processors;

a storage device to store one or more programs,

when executed by the one or more processors, cause the one or more processors to implement the child resistant method of any one of claims 1-10.

In a fourth aspect, the embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for preventing a child from getting lost as claimed in any one of claims 1 to 10.

According to the method for preventing the children from being lost, provided by the embodiment of the invention, the monitoring video stream of a target scene is obtained; determining whether a child is present in the target scene based on the surveillance video stream; if the children exist in the target scene, performing depth analysis on the monitoring video stream to obtain an analysis result; determining whether the child is at risk of getting lost based on the analysis result; and if the children have the risk of losing, alarming, and achieving the purpose of preventing all children in a large area from losing.

Drawings

Fig. 1 is a schematic flow chart of a method for preventing children from getting lost in accordance with a first embodiment of the present invention;

FIG. 2 is a flow chart of a method for preventing children from getting lost in the second embodiment of the present invention;

FIG. 3 is a schematic flow chart illustrating a method for preventing children from getting lost in a third embodiment of the present invention;

FIG. 4 is a flow chart illustrating a method for preventing children from getting lost in a fourth embodiment of the present invention;

FIG. 5 is a schematic flow chart illustrating a method for preventing children from getting lost in the fifth embodiment of the present invention;

fig. 6 is a flow chart of a method for preventing children from getting lost in the sixth embodiment of the invention;

fig. 7 is a schematic structural diagram of a child-resistant apparatus according to a seventh embodiment of the present invention;

fig. 8 is a schematic structural diagram of an electronic device in an eighth embodiment of the present invention.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not to be construed as limiting the invention. It should be further noted that, for the convenience of description, only some structures related to the present invention are shown in the drawings, not all of them.

Example one

Fig. 1 is a schematic flow chart of a method for preventing children from being lost according to an embodiment of the present invention, which is applicable to safety monitoring of children in public places with dense crowds, such as shopping malls and downtown areas, to prevent children from being lost, and the method may be executed by a device for preventing children from being lost, where the device may be implemented by software and/or hardware. Referring specifically to fig. 1, the method for preventing children from getting lost includes the following steps:

and step 110, acquiring a monitoring video stream of the target scene.

The target scene comprises large-area places such as superstores and downtown areas, and can also comprise any places such as supermarkets where children safety monitoring is needed. The monitoring video stream can be acquired by cameras arranged around a target scene, the number of the cameras can be set according to the area of the target scene, and the purpose is to realize 360-degree dead-angle-free monitoring of the target scene so as to achieve the purpose of safely monitoring children at any position in the target scene.

Step 120, determining whether a child exists in the target scene based on the monitoring video stream, and if it is determined that a child exists in the target scene, continuing to execute step 130.

Specifically, at least two continuous frames of original pictures are obtained based on the monitoring video stream, all faces in the original pictures are identified based on the original pictures in combination with a face identification technology, then age information corresponding to each face is identified and presumed in combination with the picture identification technology, and finally whether children exist in a target scene or not is determined according to the age information corresponding to each face. Or identifying the clothing type of each person in the original picture by using a picture identification technology, and determining whether a child exists in the target scene according to the clothing type, wherein the clothing type specifically comprises: children's apparel and adult apparel. The height data of each person in the original picture can be identified through a face identification technology and a picture identification technology, and whether children exist in the target scene or not is determined through the height data.

The operation of determining whether children exist in the target scene based on the monitoring video stream can be executed by monitoring terminals arranged around the target scene, and can also be executed by a cloud analysis server; if the operation is executed by the cloud analysis server, the monitoring terminal needs to upload the monitoring video stream of the target scene acquired in real time to the cloud analysis server in real time.

And step 130, performing depth analysis on the monitoring video stream to obtain an analysis result.

Wherein the essence of performing depth analysis on the surveillance video stream is: the emotional state of the child in the monitoring video stream is analyzed, for example, whether the child in the monitoring video stream has crying, anxious or panic and other emotional manifestations that can reveal that the child has a risk of losing at present is analyzed. Specifically, the child monitoring video stream with specific emotional expression can be learned in advance based on a deep learning technology to memorize various emotional expression characteristics, so that the purpose of identifying the emotional expression of the child in the newly-coming monitoring video stream is realized.

Specifically, the operation of performing the depth analysis on the surveillance video stream may be performed by a cloud analysis server, and if the operation of determining whether a child exists in the target scene based on the surveillance video stream in step 120 is performed by a surveillance terminal arranged around the target scene, when the surveillance terminal determines that a child exists in the target scene, the surveillance video stream including the child needs to be sent to the cloud analysis server, so that the cloud analysis server performs the depth analysis on the surveillance video stream.

And 140, determining whether the child has a risk of getting lost or not based on the analysis result, and if so, executing the step 150.

Whether the child has the risk of getting lost or not can be determined according to the emotional state of the child, for example, if the current emotional state of the child is worried, crying or panic and misbehavior, the current child is determined to have the risk of getting lost. In order to improve the accuracy of determining whether the child is at risk of loss, the determination may be performed based on the emotional state of the child in combination with the information of the adults around the child, for example, if the current emotional state of the child is anxious and there are no adults around the current child, the current child is determined to be at risk of loss. Or, if the current emotional state of the child is anxious, and there are adults around the current child, but the line of sight or the focus of attention of the adults is not on the current child, determining that the current child is at risk of getting lost. Or if the current emotional state of the child is anxious, and there are adults around the current child but the adults are not related to the current child, determining that the current child is at risk of getting lost.

Further, the method for preventing children from getting lost further comprises the following steps: if the child has the risk of getting lost, storing the monitoring video stream containing the target scene of the child, and sending the monitoring video stream to terminal equipment (such as a computer of a monitoring room) corresponding to security personnel for manual review, so as to improve the accuracy of determining the risk of getting lost of the monitored child, and prevent false alarm and confusion of people; meanwhile, timely storage of the lost site information of the children is achieved, if the monitored children really have the risk of losing, the lost children can be quickly searched based on the stored lost site information of the children, and the safety of the children is greatly improved.

Specifically, the operation of determining whether the child is at risk of getting lost based on the analysis result may be performed by a cloud analysis server, or may be performed by an alarm system. If the operation of determining whether the child has the loss risk based on the analysis result is executed by the cloud analysis server, when the current child is determined to have the loss risk, the cloud analysis server needs to send an alarm signal to an alarm system so that the alarm system can give an alarm in time. If the operation of determining whether the child has the risk of losing based on the analysis result is executed by the alarm system, the cloud analysis server needs to send the analysis result obtained by performing deep analysis on the monitoring video stream to the alarm system, and the alarm system performs logic judgment and alarm decision.

And 150, alarming.

If the monitored child has the risk of losing, an alarm is given to improve the attention of related guardians such as parents and relatives of the child.

According to the method for preventing the children from getting lost, the monitoring video stream of the target scene is obtained; determining whether a child is present in the target scene based on the surveillance video stream; if the children exist in the target scene, performing depth analysis on the monitoring video stream to obtain an analysis result; determining whether the child is at risk of getting lost based on the analysis result; if the children have the risk of getting lost, the technical means of alarming is carried out, the purpose of preventing all children in a large-area range from getting lost is achieved, when the children have the risk of getting lost, the monitoring video stream containing the target scene of the children is stored and sent to the terminal equipment corresponding to the security personnel, manual review is carried out, the accuracy of determining the risk of getting lost of the children is improved, meanwhile, timely storage of the information of the scene of getting lost of the children is achieved, and the children can be found as soon as possible.

Example two

Fig. 2 is a schematic flow chart of a method for preventing children from getting lost according to a second embodiment of the present invention. On the basis of the foregoing embodiment, in this embodiment, the step 120 of determining whether a child exists in the target scene based on the monitoring video stream is optimized, specifically, the identification of the child in the target scene is realized by combining a face recognition technology and an image recognition technology. Referring specifically to fig. 2, the method comprises the following steps:

and step 210, acquiring a monitoring video stream of a target scene.

Step 220, extracting at least one frame of original picture from the monitoring video stream.

And step 230, for each frame of original picture, inputting the current frame of original picture into a preset face detection model to obtain coordinate position information and size information of each face in the current frame of original picture.

Specifically, each frame of original picture is sequentially input into a preset face detection model, the face detection model performs operation on each frame of original picture to obtain coordinate position information and size information of each face, wherein the coordinate position information of each face specifically refers to coordinate position information of key points of the face, such as coordinate position information of two eyes, coordinate position information of a nose and coordinate position information of a mouth, and the size information specifically refers to size information of a face contour.

The face detection model is constructed based on a deep learning algorithm and is obtained by pre-training a large number of labeled face pictures, wherein the labeled face pictures refer to pictures of face coordinate position information and size information in the labeled pictures, so that the face detection model learns and memorizes an operation process for recognizing face characteristic information, and the effect of recognizing the coordinate position information and the size information of faces in various pictures is achieved. When a picture is sent into the trained face detection model, the face detection model automatically detects whether a face exists in the picture, if the face exists in the picture, the coordinate position information and the size information of the face are further detected, and the detected coordinate position information and the detected size information of the face are output.

And step 240, for the coordinate position information and the size information of each face, intercepting a local picture containing the current face from the corresponding frame original picture according to the coordinate position information and the size information of the current face.

For the coordinate position information and the size information of each face, the essence of intercepting the local picture containing the current face from the corresponding frame original picture according to the coordinate position information and the size information of the current face is as follows: and respectively intercepting local pictures from corresponding frame original pictures based on the coordinate position information and the size information of each face, namely each local picture comprises one face, and the corresponding frame original pictures refer to pictures containing the faces.

Each original picture is represented by a large number of coordinate data points, and when coordinate position information and size information of the current face are obtained, data points belonging to the current face can be taken out from the large number of coordinate data points, so that the process of intercepting a local picture containing the current face from the corresponding frame of original picture is realized.

Step 250, inputting each local picture into a preset age identification model to obtain an age numerical value corresponding to the face in each local picture.

Inputting each local picture into a preset age identification model, and obtaining the essence of the age numerical value corresponding to the face in each local picture as follows: and sequentially inputting each local picture into a preset age identification model, and respectively calculating each local picture by the age identification model to obtain the age numerical value of the face in each local picture.

The preset age identification model is constructed based on a deep learning algorithm and is obtained by training a large number of labeled face pictures, wherein the labeled face pictures refer to pictures of age data corresponding to faces in the labeled face pictures. In order to improve the recognition accuracy of the model, different types of face pictures can be collected, for example, face pictures including faces of different ages, face pictures including multiple faces, face pictures under different light rays, and the like are collected. When a new face picture, namely the local picture, is input into the age identification model, the age identification model outputs an age value corresponding to the face in the local picture after operation.

Step 260, determining whether children exist in the target scene based on the age value, and if it is determined that children exist in the target scene, executing step 270.

Specifically, if an age value smaller than a preset value exists, it is determined that a child exists in the target scene, and optionally, the preset value may be 10, for example.

And 270, performing depth analysis on the monitoring video stream to obtain an analysis result.

And step 280, determining whether the child has the risk of losing based on the analysis result, and alarming if the child has the risk of losing.

On the basis of the technical solution of the embodiment, in the technical solution of this embodiment, at least one frame of original picture is extracted from the surveillance video stream, for each frame of original picture, the current frame of original picture is input to a preset face detection model, coordinate position information and size information of each face in the current frame of original picture are obtained, for the coordinate position information and size information of each face, a partial picture including the current face is intercepted from a corresponding frame of original picture according to the coordinate position information and size information of the current face, each partial picture is input to a preset age identification model, an age value corresponding to the face in each partial picture is obtained, and whether children exist in the target scene is determined based on the age value, so that the purpose of determining whether children exist in the target scene is achieved.

EXAMPLE III

Fig. 3 is a schematic flow chart of a method for preventing children from getting lost according to a third embodiment of the present invention. On the basis of the foregoing embodiment, this embodiment continues to optimize "determining whether there is a child in the target scene based on the surveillance video stream" in step 120, and provides another implementation manner, specifically, implementing identification of a child in the target scene by identifying the clothing category of the portrait in the picture. With particular reference to fig. 3, the method comprises the following steps:

and step 310, acquiring a monitoring video stream of the target scene.

Step 320, extracting at least one frame of original picture from the monitoring video stream.

And 330, inputting the current frame original picture into a preset matting model for each frame original picture to obtain a clothes sub-picture of each portrait clothes area in the current frame original picture.

Specifically, each frame of original picture is sequentially input into a preset cutout model, the cutout model is operated according to each frame of original picture to obtain a clothes character picture of each portrait and clothes area contained in each frame of original picture, wherein each clothes sub-picture comprises a clothes area of a portrait.

The matting model is constructed based on a deep learning algorithm and obtained by utilizing a large number of labeled portrait pictures to perform pre-training, wherein the labeled portrait pictures refer to pictures of portrait clothes regions in the labeled pictures, so that the matting model learns and memorizes the operation process of recognizing the portrait clothes regions in the pictures, and the effect of recognizing the portrait clothes regions in various pictures is achieved. When a picture is sent into the trained matting model, the matting model automatically detects a portrait clothing region in the picture, and outputs the detected portrait clothing region to obtain a clothing sub-picture comprising the portrait clothing region.

And 340, inputting the current clothes sub-picture into a preset clothes category identification model for each clothes sub-picture to obtain the category of clothes in the current clothes sub-picture.

For each clothes sub-picture, inputting the current clothes sub-picture into a preset clothes category identification model, and obtaining the essence of the category of clothes in the current clothes sub-picture is as follows: and sequentially inputting each clothes sub-picture into a preset clothes category identification model, and respectively calculating each clothes sub-picture by the clothes category identification model to obtain the category of clothes in each clothes sub-picture, wherein the category of the clothes comprises children clothes and adult clothes.

The preset clothing category identification model is constructed based on a deep learning algorithm and is obtained by training a large number of marked clothing pictures, wherein the marked clothing pictures refer to pictures of clothing categories in the marked clothing pictures. In order to improve the recognition accuracy of the model, different types of clothes pictures can be collected, for example, pictures of clothes worn by people of different ages, pictures of clothes containing clothes of different styles, pictures of clothes under different light rays, and the like can be collected. When a new clothes picture, namely the clothes sub-picture, is input into the clothes category identification model, the clothes category identification model outputs whether the category of clothes in the clothes sub-picture is adult clothes or children clothes after operation.

Step 350, determining whether children exist in the target scene based on the clothes type, and if so, executing step 360.

Specifically, if the clothing type identification model identifies that the clothing types in the clothing sub-pictures are all adult clothing, it is determined that no child exists in the target scene, and if the clothing type identification model identifies that child clothing exists in the clothing types in the clothing sub-pictures, it is determined that a child exists in the target scene.

And 360, performing depth analysis on the monitoring video stream to obtain an analysis result.

And step 370, determining whether the child has the risk of losing based on the analysis result, and alarming if the child has the risk of losing.

On the basis of the technical solution of the above embodiment, in the technical solution of this embodiment, at least one original picture is extracted from the monitoring video stream; for each frame of original picture, inputting the current frame of original picture into a preset cutout model to obtain a clothes sub-picture of each portrait clothes area in the current frame of original picture; for each clothes sub-picture, inputting the current clothes sub-picture into a preset clothes category identification model to obtain the category of clothes in the current clothes sub-picture; the technical means for determining whether children exist in the target scene based on the clothes category achieves the purpose of determining whether children exist in the target scene.

Example four

Fig. 4 is a schematic flow chart of a method for preventing children from getting lost according to a fourth embodiment of the present invention. On the basis of the foregoing embodiment, this embodiment continues to optimize "determining whether a child exists in the target scene based on the surveillance video stream" in step 120, and provides yet another implementation manner, specifically, on the basis of determining that a child exists in the target scene by combining a face recognition technology and an image recognition technology, it is determined whether the determined child is really a child through clothing category recognition, so that the optimization has the advantage of improving the accuracy of child recognition. With particular reference to fig. 4, the method comprises the following steps:

and step 410, acquiring a monitoring video stream of the target scene.

Step 420, extracting at least one original picture from the surveillance video stream.

And step 430, for each frame of original picture, inputting the current frame of original picture into a preset face detection model to obtain coordinate position information and size information of each face in the current frame of original picture.

And step 440, for the coordinate position information and the size information of each face, intercepting a local picture containing the current face from the corresponding frame of original picture according to the coordinate position information and the size information of the current face.

Step 450, inputting each local picture into a preset age identification model to obtain an age numerical value corresponding to the face in each local picture.

Step 460, determining whether children exist in the target scene based on the age value, and if it is determined that children exist in the target scene, continuing to execute step 470.

And 470, marking the portrait corresponding to the child and the portrait corresponding to the adult in the original picture respectively based on the age value corresponding to the face in each local picture.

Specifically, for example, if the age value corresponding to the face in the current partial picture is greater than a preset value, for example, 15, the person image corresponding to the current partial picture in the original picture is marked as an adult. And if the age value corresponding to the face in the current local picture is smaller than a preset value, marking the portrait corresponding to the current local picture in the original picture as a child.

Step 480, acquiring a clothes sub-picture of the marked child portrait clothes area in the original picture.

And 490, inputting the clothes sub-picture into a preset clothes category identification model to obtain the category of clothes in the clothes sub-picture.

Step 4100, determining whether the marked child figure is a real child figure based on the type of the clothing, and if the marked child figure is a real child figure, continuing to execute step 4110.

If the clothes category marked as the child figure is the child clothes, the marked child figure is determined to be the real child figure, and if the clothes category marked as the child figure is the adult clothes, the marked child figure is determined not to be the real child figure.

Step 4110, performing depth analysis on the monitoring video stream to obtain an analysis result.

And 4120, determining whether the child has the risk of losing based on the analysis result, and alarming if the child has the risk of losing.

On the basis of the technical scheme of the embodiment, according to the technical scheme of the embodiment, on the basis that the face recognition technology and the picture recognition technology are combined to determine that the child exists in the target scene, whether the determined child is really the child is determined through the clothing category recognition technology, and therefore the accuracy of identifying the child in the target scene is improved.

EXAMPLE five

Fig. 5 is a schematic flow chart of a method for preventing children from getting lost according to a fifth embodiment of the present invention. On the basis of the foregoing embodiment, in this embodiment, if it is determined that there is a child in the target scene in step 130, "deep analysis is performed on the monitoring video stream to obtain an analysis result," optimization is performed, and specifically, recognition of the emotional state of the face in the monitoring video stream is achieved by combining with a deep learning technique. Referring specifically to fig. 5, the method comprises the steps of:

and step 510, acquiring a monitoring video stream of the target scene.

Step 520, determining whether children exist in the target scene based on the monitoring video stream, and if so, executing step 530.

Step 530, at least two continuous original pictures are extracted from the monitoring video stream.

And 540, for each frame of original picture, identifying the outline coordinate information of each preset child face organ in the current frame of original picture by a face identification key point extraction technology.

Specifically, the face recognition technology is used for recognizing the face in each frame of original picture, then the child face in the face is recognized based on the picture recognition technology, and finally the outline coordinate information of preset organs of each child face is recognized through the face recognition key point extraction technology, wherein the preset organs comprise eyes, a mouth, a nose and the like.

And 550, connecting the outline coordinate information of the preset organs of the faces of the children in the original pictures of the frames in series according to the time sequence relation.

Specifically, according to the time sequence, the contour coordinate information of the preset organs of the same child face in each frame of original picture is connected in series, and then the information corresponding to each child after being connected in series is input into a preset emotion recognition model, so that the emotion state corresponding to each child face is obtained, wherein the emotion state comprises crying and anxiety. By analyzing the change of the outline coordinate information of the preset organs of the face of the child in the multi-frame original pictures in a continuous period of time, the emotional state analysis result with higher accuracy can be obtained.

And 560, inputting the information after the series connection into a preset emotion recognition model to obtain the emotion state corresponding to each child face.

Specifically, the information after the series connection corresponding to each child face is input into a preset emotion recognition model, and an emotion state corresponding to each child face is obtained. The preset emotion recognition model is constructed based on a deep learning algorithm and is obtained by training a large number of labeled human face picture sequence frames, wherein the labeled human face picture sequence frames refer to continuous multi-frame pictures of emotion states corresponding to a labeled human face. In order to improve the recognition accuracy of the model, the face picture sequence frames in different emotional states can be collected, for example, the face picture sequence frame of a child whose emotional state is anxious, the face picture sequence frame of a child whose emotional state is crying, and the like are collected. When a new face picture sequence frame is input into the emotion recognition model, the emotion recognition model outputs the emotion state corresponding to the face in the face picture sequence frame after operation.

And 570, when the current emotional state of the child is crying or anxious and no adult exists in the preset distance range from the current child, determining that the current child has a risk of getting lost, and giving an alarm.

On the basis of the technical scheme of the embodiment, the technical scheme of the embodiment identifies the emotional state of the face by using the continuous multi-frame original pictures, improves the identification precision of the emotional state of the face, is beneficial to determining whether the monitored child has a loss risk, and realizes the loss prevention monitoring of the child in a large-area target scene.

EXAMPLE six

Fig. 6 is a schematic flow chart of a method for preventing children from getting lost according to a sixth embodiment of the present invention. On the basis of the foregoing embodiment, if it is determined that a child exists in the target scene in step 130, "in this embodiment, the monitoring video stream is subjected to deep analysis, and an analysis result" is obtained, "which is optimized continuously, specifically, the similarity between the faces of the adult and the child in the picture is identified by using a picture identification technology, so as to determine whether a blood relationship exists between the adult and the child, and to assist in determining whether the child has a risk of getting lost, so that the accuracy of determining the risk of getting lost of the child is improved. Referring specifically to fig. 6, the method includes the steps of:

and step 610, acquiring a monitoring video stream of the target scene.

Step 620, determining whether children exist in the target scene based on the monitoring video stream, and if so, executing step 630.

Step 630, at least two consecutive original pictures are extracted from the surveillance video stream.

And step 640, identifying the outline coordinate information of each preset child face organ in the current frame original picture by using a face identification key point extraction technology for each frame original picture.

And 650, connecting the outline coordinate information of the preset organs of the faces of the children in the original pictures of the frames in series according to the time sequence relation.

And 660, inputting the information after series connection into a preset emotion recognition model to obtain the emotion state corresponding to each child face.

Wherein said emotional state comprises crying and anxiety.

And step 670, determining the similarity between the adult face and the child face in the current frame original picture for each frame original picture.

Specifically, a similarity calculation model can be constructed through a deep learning technology, and the similarity between the aged face and the child face in the original picture is calculated through the similarity calculation model. The similarity calculation model can be trained by a large number of marked pictures, the marked pictures refer to pictures with similarity between faces in the marked pictures, and the process of calculating the similarity between the faces in the pictures is learned and memorized by training the similarity calculation model. When a new picture is input into the trained similarity calculation model, the model automatically calculates the similarity and outputs a similarity result.

And step 680, determining whether a blood relationship exists between the adult and the child according to the similarity.

Specifically, if the similarity is greater than a set threshold, the existence of a blood-related relationship between the adult and the child is determined, and the threshold may be 80%, for example.

And 690, when the current emotional state of the child is crying or anxious, and an adult exists in a preset distance range from the current child, but a bloody relationship does not exist between the adult and the current child, determining that the current child has a risk of losing, and giving an alarm.

On the basis of the technical scheme of the embodiment, the technical scheme of the embodiment judges whether relatives having a blood relationship with the child exist around the child by further calculating the face similarity between the child and adults around the child, further assists in judging whether the child has a risk of losing, improves the judgment precision of the risk of losing, and realizes the safety monitoring of the child in a target scene.

EXAMPLE seven

Fig. 7 is a schematic structural diagram of a child-resistant device according to a seventh embodiment of the present invention. Referring to fig. 7, the apparatus includes: an acquisition module 710, a child determination module 720, an analysis module 730, a risk determination module 740, and an alarm module 750;

the acquiring module 710 is configured to acquire a surveillance video stream of a target scene; a child determining module 720, configured to determine whether a child exists in the target scene based on the surveillance video stream; the analysis module 730 is configured to perform depth analysis on the monitoring video stream to obtain an analysis result if it is determined that a child exists in the target scene; a risk determination module 740 for determining whether the child is at risk of getting lost based on the analysis result; and the alarm module 750 is used for giving an alarm if the child has the risk of losing.

Further, the child determination module 720 includes: the extraction unit is used for extracting at least one frame of original picture from the monitoring video stream; the face recognition unit is used for inputting the current frame original picture into a preset face detection model for each frame original picture to obtain coordinate position information and size information of each face in the current frame original picture; the intercepting unit is used for intercepting a local picture containing the current face from a corresponding frame original picture according to the coordinate position information and the size information of the current face for the coordinate position information and the size information of each face; the age identification unit is used for inputting each local picture into a preset age identification model to obtain an age numerical value corresponding to the face in each local picture; a determining unit for determining whether a child is present in the target scene based on the age value.

Further, the apparatus further comprises: and the marking module is used for marking the portrait corresponding to the child and the portrait corresponding to the adult in the original picture respectively based on the age values corresponding to the faces in the local pictures when the children exist in the target scene.

Further, the child determination module 720 further includes: the clothes sub-picture acquiring unit is used for acquiring a clothes sub-picture of a marked child portrait clothes area in the original picture when the child is determined to exist in the target scene; the clothes type identification unit is used for inputting the clothes sub-picture into a preset clothes type identification model to obtain the type of clothes in the clothes sub-picture; a confirmation unit for confirming whether the marked child figure is a real child figure based on the category of the clothes; wherein the categories of clothing include children's garments and adult garments.

Further, the child determination module 720 further includes: the matting module is used for inputting the current frame original picture into a preset matting model for each frame original picture to obtain a clothes sub-picture of each portrait clothes area in the current frame original picture; a determination unit for determining whether a child is present in the target scene based on the category of the clothing.

Further, the analysis module 730 includes: the picture extraction unit is used for extracting at least two continuous original pictures from the monitoring video stream; the face key point identification unit is used for identifying contour coordinate information of each preset child face organ in the current frame original picture through a face identification key point extraction technology for each frame original picture; the series unit is used for connecting the outline coordinate information of the preset organs of the faces of the children in the original pictures of the frames in series according to the time sequence relation; the input unit is used for inputting the information after the series connection to a preset emotion recognition model to obtain an emotion state corresponding to each child face; wherein the emotional state comprises crying and anxiety.

Further, the risk determination module 740 is specifically configured to: when the current emotional state of the child is crying or anxious, and no adult exists in the preset distance range from the current child, it is determined that the current child has a risk of getting lost.

Further, the analysis module 730 further includes: the similarity determining unit is used for determining the similarity between the adult face and the child face in the current frame original picture for each frame original picture; and the blood relationship determining unit is used for determining whether blood relationship exists between the adult and the child according to the similarity. Correspondingly, the risk determining module 740 is specifically configured to: and when the current emotional state of the child is crying or anxious, and an adult exists in a preset distance range from the current child, but the adult does not have a blood relationship with the current child, determining that the current child has a risk of getting lost.

Further, the device further comprises a storage module, which is used for storing the monitoring video stream of the target scene and sending the monitoring video stream to a terminal device corresponding to a security worker for manual review if the child has a risk of losing.

The device for preventing children from getting lost provided by the embodiment obtains the monitoring video stream of the target scene; determining whether a child is present in the target scene based on the surveillance video stream; if the children exist in the target scene, performing depth analysis on the monitoring video stream to obtain an analysis result; determining whether the child is at risk of getting lost based on the analysis results; if the children have the risk of losing, the technical means of alarming is carried out, the purpose of monitoring all children in a large-area range in a loss prevention mode is achieved, when the children have the risk of losing, the monitoring video stream containing the target scene of the children is stored and sent to the terminal equipment corresponding to the security personnel to be manually checked, the accuracy of determining the risk of losing the children is improved, meanwhile, timely storage of the information of the scene of losing the children is achieved, and the children can be found as soon as possible.

The child wandering prevention apparatus provided by the embodiment of the present invention may execute the child wandering prevention method provided by any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

Example eight

Fig. 8 is a schematic structural diagram of an electronic device according to an eighth embodiment of the present invention. FIG. 8 illustrates a block diagram of an exemplary device 12 suitable for use in implementing embodiments of the present invention. The device 12 shown in fig. 8 is only an example and should not impose any limitation on the functionality and scope of use of embodiments of the present invention.

As shown in FIG. 8, device 12 is in the form of a general purpose computing device. The components of device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including the system memory 28 and the processing unit 16.

Bus 18 represents one or more of any of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, and a processor or local bus using any of a variety of bus architectures. By way of example, such architectures include, but are not limited to, industry Standard Architecture (ISA) bus, micro-channel architecture (MAC) bus, enhanced ISA bus, video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

Device 12 typically includes a variety of computer system readable media. Such media may be any available media that is accessible by device 12 and includes both volatile and nonvolatile media, removable and non-removable media.

The system memory 28 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM) 30 and/or cache memory 32. Device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 8 and commonly referred to as a "hard drive"). Although not shown in FIG. 8, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 by one or more data media interfaces. Memory 28 may include at least one program product having a set of program modules (e.g., acquisition module 710, child determination module 720, analysis module 730, risk determination module 740, and alarm module 750 of the child loss prevention device) configured to perform the functions of embodiments of the present invention.

A program/utility 40 having a set of program modules 42 (e.g., an acquisition module 710, a child determination module 720, an analysis module 730, a risk determination module 740, and an alarm module 750 of the child-resistant device) may be stored, for example, in memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may include implementation of a network environment. Program modules 42 generally carry out the functions and/or methodologies of the described embodiments of the invention.

Device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), with one or more devices that enable a user to interact with device 12, and/or with any devices (e.g., network card, modem, etc.) that enable device 12 to communicate with one or more other computing devices. Such communication may be through an input/output (I/O) interface 22. Also, the device 12 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the Internet) via the network adapter 20. As shown, the network adapter 20 communicates with the other modules of the device 12 via the bus 18. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, to name a few.

The processing unit 16 executes various functional applications and data processing by executing programs stored in the system memory 28, for example, to implement the child-resistant method provided by the embodiment of the present invention.

Example nine

An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for preventing a child from getting lost, where the method includes:

acquiring a monitoring video stream of a target scene;

determining whether the child is at risk of getting lost based on the analysis result;

and if the child has the risk of losing, giving an alarm.

Computer storage media for embodiments of the invention may employ any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).

It is to be noted that the foregoing description is only exemplary of the invention and that the principles of the technology may be employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method of preventing a child from becoming lost, comprising:

acquiring a monitoring video stream of a target scene;

acquiring at least two continuous original pictures based on the monitoring video stream, and determining whether children exist in the target scene based on the original pictures;

if the child has the risk of losing, alarming;

wherein, the performing depth analysis on the monitoring video stream to obtain an analysis result includes:

for each frame of original pictures in at least two continuous original pictures in the monitoring video stream, identifying outline coordinate information of each child face preset organ in the current frame of original pictures by a face identification key point extraction technology;

connecting outline coordinate information of each child face preset organ in each frame of original picture in series according to a time sequence relation;

inputting the information after series connection into a preset emotion recognition model to obtain an emotion state corresponding to each child face;

wherein the emotional state comprises crying and anxiety.

2. The method of claim 1, wherein determining whether a child is present in the target scene based on the surveillance video stream comprises:

extracting at least one original picture from the monitoring video stream;

for each frame of original picture, inputting the current frame of original picture into a preset face detection model to obtain coordinate position information and size information of each face in the current frame of original picture;

for the coordinate position information and the size information of each face, intercepting a local picture containing the current face from a corresponding frame original picture according to the coordinate position information and the size information of the current face;

inputting each local picture into a preset age identification model to obtain an age numerical value corresponding to the face in each local picture;

determining whether a child is present in the target scene based on the age value.

3. The method of claim 2, when it is determined that a child is present in the target scene, further comprising:

and respectively marking the portrait corresponding to the child and the portrait corresponding to the adult in the original picture based on the age value corresponding to the face in each local picture.

4. The method of claim 3, when it is determined that a child is present in the target scene, further comprising:

acquiring a clothes sub-picture of a marked child portrait clothes area in an original picture;

inputting the clothes sub-picture into a preset clothes category identification model to obtain the category of clothes in the clothes sub-picture;

confirming whether the marked child figure is a real child figure based on the category of the clothing;

wherein the categories of clothing include children's garments and adult garments.

5. The method of claim 1, wherein determining whether a child is present in the target scene based on the surveillance video stream comprises:

extracting at least one original picture from the monitoring video stream;

for each frame of original picture, inputting the current frame of original picture into a preset cutout model to obtain a clothes sub-picture of each portrait clothes area in the current frame of original picture;

for each clothes sub-picture, inputting the current clothes sub-picture into a preset clothes category identification model to obtain the category of clothes in the current clothes sub-picture;

determining whether a child is present in the target scene based on the category of the clothing.

6. The method of claim 1, wherein determining whether the child is at risk of getting lost based on the analysis comprises:

when the current emotional state of the child is crying or anxious, and no adult exists in the preset distance range from the current child, it is determined that the current child has a risk of getting lost.

7. The method of claim 1, wherein performing a depth analysis on the surveillance video stream to obtain an analysis result further comprises:

for each frame of original picture, determining the similarity between the adult face and the child face in the current frame of original picture;

and determining whether a blood relationship exists between the adult and the child according to the similarity.

8. The method of claim 7, wherein determining whether the child is at risk of getting lost based on the analysis comprises:

and when the current emotional state of the child is crying or anxious, and an adult exists in a preset distance range from the current child, but the adult does not have a blood relationship with the current child, determining that the current child has a risk of getting lost.

9. The method of any one of claims 1-5, further comprising: and if the child has the risk of losing, storing the monitoring video stream of the target scene, and sending the monitoring video stream to terminal equipment corresponding to security personnel for manual rechecking.

10. A child-resistant device comprising:

the child determination module is used for acquiring at least two continuous original pictures based on the monitoring video stream and determining whether a child exists in the target scene based on the original pictures;

the analysis module is used for identifying contour coordinate information of each child face preset organ in the current frame original picture through a face identification key point extraction technology for each frame of at least two continuous original pictures in the monitoring video stream if the child exists in the target scene; connecting outline coordinate information of each preset child face organ in each frame of original picture in series according to a time sequence relation; inputting the information after the series connection into a preset emotion recognition model to obtain an emotion state corresponding to each child face; wherein the emotional state comprises crying and anxiety; a risk determination module for determining whether the child is at risk of getting lost based on the analysis result;

and the alarm module is used for giving an alarm if the child has a risk of getting lost.

11. An electronic device, characterized in that the electronic device comprises:

one or more processors;

a storage device for storing one or more programs,

when executed by the one or more processors, cause the one or more processors to implement the child resistant method of any one of claims 1-9.

12. A computer-readable storage medium, on which a computer program is stored, which program, when being executed by a processor, carries out a method of preventing a child from wandering as set forth in any one of claims 1-9.